Language Models & Methods

AI boom

Development (sth about the Cold War)

AI winter

Down moment (apparently, we’re entering our third AI winter rn?)

N-grams

Predict the next word

Language Models

Tools that process and generate human-like text based on patterns in data. They understand text (how to classify or extract information), and generate text (ability to summarize, translate or answer questions). EX: chatbots, search engines

Why are LMs important?

Automation, accessibility and Efficiency.

What gives LMs their high efficiency?

Ability the process massive amounts of text quickly

LM’s automation?

Tasks like summarizing reports and answering FAQas

How come are LMs accessible?

They’re able to make complex text easier to understand

ELIZA (what is it + date)

Rule based system that relied on hard coded grammar rules and syntax. 1964

Trad models

Relied on rules, keywords, statistical patterns. Limited.

Modern models

AAE

Diffference trad and modern?

Trad is limited, no awareness of the context provided. Modern processes globally.

Where could Text Data & LM be applied?

Libraries and Archives, Policy and Governance, IS (deep dive on Notion and BS slides W1)